In this paper, we firstly introduce an approach to the modeling of a domain-specific ontology for use in connection with a given document collection. Secondly, we present a methodology for deriving conceptual similarity from the domain-specific ontology. Adopted for ontology representation is a specific lattice-based concept algebraic language by which ontologies are inherently generative. The modeling of a domain specific ontology is based on a general ontology built upon common knowledge resources as dictionaries and thesauri. Based on analysis of concept occurrences in the object document collection the general ontology is restricted to a domain specific ontology encompassing concepts instantiated in the collection. The resulting domain specific ontology and similarity can be applied for surveying the collection through key concepts and conceptual relations and provides a means for topic-based navigation. Finally, a measure of concept similarity is derived from the domain specific ontology based on occurrences, commonalities, and distances in the ontology.
|Titel||Lecture Notes in Artificial Intelligence|
|Status||Udgivet - 2005|
|Begivenhed||15th International Symposium on Methodologies for Intelligent Systems, ISMIS 2005 - Saratoga Springs, USA|
Varighed: 25 maj 2005 → 28 maj 2005
Konferencens nummer: 15
|Konference||15th International Symposium on Methodologies for Intelligent Systems, ISMIS 2005|
|Periode||25/05/2005 → 28/05/2005|